Statement of cloud spend
One statement.
Every cloud bill.
Capitops pulls your cloud (AWS, Azure, GCP, Oracle Cloud), AI (OpenAI, Anthropic), and SaaS spend — GitHub included, and even per-seat subscriptions that have no billing API at all — into one normalized ledger, down to the token. Every number traces to its line items; the savings hiding in it get flagged. Built on FOCUS, hosted entirely in Canada.
Google or work email — no credit card.
Example reconciliation
- AWS
- $18,402.11
- Azure
- $9,110.40
- GCP
- $4,988.72
- AI APIs(OpenAI · Anthropic)
- $1,206.33
- Seats & subscriptions(GitHub · per-seat AI)
- $612.00
What's on the statement
Every source, one row shape
Provider, resource, region, tags, reconciled cost — normalized to FOCUS, the FinOps Foundation's open billing spec, so a charge from any source reads the same way.
FinOps for AI, down to the token
Cost per model, $/1M tokens, and cache-hit rate for every AI/LLM connection — the consumption-level detail that turns “AI spend went up” into “why.”
Every number traceable
Click a service and follow the dollars down — to the resources behind it, then to the underlying FOCUS line items. No figure on the dashboard is a dead end.
Savings, found for you
Unused seats, renewals about to auto-bill, rightsizing your providers already recommend — collected on one page, each with a link into the provider's own console to act on it.
Synced automatically
Fresh spend every 6 hours, up to a year of history backfilled when you connect, and recently closed months re-checked so provider restatements don't slip past.
Sliced your way
Spend by account, by tag, billed or amortized, over any date range — and any view exports as FOCUS-shaped CSV when you need it somewhere else.
The month, in plain English
Slack budget and spike alerts, a month-end forecast, and an opt-in AI-written monthly digest — the numbers come to you, no dashboard habit required.
Stored in Canada
Your database lives in Montreal. Provider APIs are read where they live — every cross-border read is disclosed in full, per source, before you connect it.